Long Short-term Memory Model for Temperature Forecasting in Khartoum, Sudan

Abstract

This ponder presents LSTM model to forecast temperature trends for Khartoum Sudan utilizing  (CRU) data The dataset spans multiple a long time and has been split into 80 for training and 20 for testing to guarantee strong model evaluation The LSTM model successfully captures complex transient conditions and seasonality patterns within the historical temperature data The results demonstrate a sensible predictive performance with the model appearing an increasing temperature trend up to the year 2026 The discoveries suggest that deep learning models particularly LSTM are appropriate for long-term climate forecasting in bone-dry regions such as Khartoum where temperature changes are significant.